Record Luna Max retrieval outcome

[ashwingopalsamy]

Summary

• Record the 32-run GPT-6 Luna Max Q&A availability screen against native Go tools.
• Add the blind-review scores, source and prompt hashes, token and latency summaries, and prompted-use diagnostics.
• Update the research note and continuation handoff: current retrieval did not show natural Q&A value, so persistent indexing and the separate accepted-change screen remain gated.

Validation

• go test ./...
• go test -race ./...
• go vet ./...
• go build ./...
• git diff --check
• Parsed the result manifest as JSON.